With the Alt-A market growing on the back of a strong lender push, analysts have been examining how to divvy up the market to more accurately assess risk.
"Currently there is a lot of discussion about differences in the types of product that fall between prime and subprime," said Satish Mansukhani, head of mortgage strategy at Credit Suisse First Boston, adding that there is a borrower base out there that is not served by either the prime or subprime markets.
Rating agencies also agree that differentiating Alt-A collateral is crucial in understanding the market. "The recent research we are seeing on the Alt-A sector from various market participants, all seem to agree in concept that the Alt-A market should be segmented in different tiers," said Sarbashis Ghosh, a senior director from Fitch Ratings. He added that the drivers of the segmentation are similar - the only point of contention is which of these have a stronger impact. In other words, it's a difference in opinion on what the major causes for the segmentation are, not whether the segmentation in Alt-As actually exists.
Ghosh explained that one must look at the quantitative (which could be derived from the data) and qualitative (which varies from issuer to issuer) characteristics of an Alt-A mortgage. He explained that even though a loan might have similar attributes, if the issuer is different, the performance might vary. "Loan performance is driven by factors other than what the data represents," he said.
In a report released in October, Fitch explained that in a nonhomogenous product like Alt-A, lender-specific intangibles (eg., underwriting process and controls or valuation procedures) play an important role in discerning differences in embedded risks and credit performance. The rating agency explained that these intangibles manifest themselves through performance measures including delinquency status and pool losses, even though there are no obvious differences in the underlying risk characteristics of the collateral pool (see ASR 10/18/04).
CSFB in a recent report also said that the multiple non-standard features in Alt-A collateral leads to the origination of mortgages with layered risk. Analysts stated that they are finding a distinct correlation between these layers and the rate premium - the differential between the rate on a loan and the benchmark conforming rate. Analysts, therefore, view this measure as a good way to differentiate loans in the Alt-A market. "We prefer the use of a quantifiable metric like rate premium to classify the various grades of Alt-A product over the use of subjective and potentially fuzzy labels," wrote analysts.
CSFB said that the compensation for the non-standard features in Alt-A collateral is reflected in a higher rate to borrower, which is evidenced by the rate premium on these loans. Analysts also found that an increase in the rate premium is proportional to the decline in the lender's assessment of borrower and loan features relating to capacity, collateral and credit, adding that there is a correlation between rate premiums and changes in certain borrower features. For instance, correlations exist between the rate premium and a general decline in the average FICO, a notable rise in the number of loans with FICO higher than 640, and a significant increase in the number of loans with less than prime features, such as no documentation mortgages, purchase-money loans, as well as loans with LTV ratios over 90%.
A further differentiation of the sector would be to distinguish between the characteristics of conforming versus non-conforming Alt-A loans. CSFB said that the inclusion of loans into non-conforming and conforming Alt-As is predominantly driven by the exclusion of these loans from prime jumbo pools and the GSE "box."
"We believe that the characteristics of non-conforming Alt-As should be benchmarked to prime jumbos and those of conforming-balance Alt-As to Agency pools backed by prime mortgages," analysts wrote, adding that conforming Alt-A loans, in aggregate, show a greater concentration of investor properties and loans with original LTV over 80% versus non-conforming Alt-As.
Separately, Fitch held a teleconference last Tuesday presenting its analysis on the tiering of Alt-A. Expanding on the October research report, analysts discussed additional delinquency performance metrics, loss expectations and the issuers in the different tiers. Fitch also discussed loan performance based on 90 plus day delinquencies and observed similar trends to that of 60 plus day delinquencies, which analysts covered in the earlier report. The rating agency also presented its sensitivity analysis on losses for expected and stressed scenarios, providing an indication of the extent of differences in the cumulative losses seen in pools from various Alt-A segments. They also looked at the variation in the losses as economic conditions deteriorate, reflected by stressed defaults and higher severities.
Copyright 2004 Thomson Media Inc. All Rights Reserved.